FAQ

Cross-track common questions. Sorted by audience role — find your role and jump. For capability-specific issues, see Common pitfalls.


Overview

You are Jump to
Writer / teacher / blogger §1 Writers
AI app / prompt engineer §2 AI engineers
Integration developer (using @rho/md) §3 Integrators
Business procurement / legal §4 Business / legal
Protocol implementer (writing your own reader) §5 Protocol implementers
Academic researcher §6 Academic

1. Writers

Q: I'm a blogger / teacher / student — where do I start?

→ What is Rho? — 5-min philosophy → 5-minute tour — see all 15 capabilities → Your first interactive .md — build a compound interest calculator → Cheatsheet — printable one-page reference

Q: Where can I render my Rho documents?

In Rho MD — the free app for Windows, Linux and Android (download). There is no browser editor; rendering happens in the app. The renderer also exists as a library for embedding in your own site, but it is not published to npm yet.

See Install Rho app.

Q: Can my .md file be viewed on GitHub?

✅ Yes — Rho protocol guarantees plain-text fallback. GitHub web rendering: standard markdown parts render normally + Rho blocks display as code-block literals. Readers can still understand the content fully.

Q: I don't want to install anything — can I use it directly?

To write, yes — any text editor works, and the file reads as ordinary Markdown anywhere (GitHub, a gist, your notes app). To see it run, you need the Rho MD app; there is no browser version.

Q: My personal blog uses Rho — free?

✅ Fully free. See Free use FAQ.

Q: I run paid online courses — can I use Rho?

⚠️ Commercial use requires a commercial license. Self-implementing from spec is fully free. See Free use FAQ + Commercial licensing.

Q: How do I avoid common mistakes?

→ Common pitfalls — cross-capability overview

Q: I want to see more demos for ideas

→ Examples gallery — 20+ complete copy-paste demos


2. AI engineers

Q: I'm building an AI app — how do I get LLMs to generate Rho?

→ AI/LLM Guide overview — 3 integration modes → Prompt templates — 7 ready-to-use prompts → Schema-aware generation — high-accuracy function calling → Common LLM mistakes & repair — 10 LLM errors + auto-repair

Q: Which mode has the highest accuracy?

Raw prompt:                ~80%
Few-shot examples:         ~90%
Function calling / JSON:   ~99%
Fine-tuning:               99.5%+

For production, recommended: function calling + 5-10 few-shot examples + server-side validation.

Q: Can LLM output render directly to end users? Is it safe?

✅ Safe. Rho protocol-layer forbids arbitrary JS / network / eval / side effects. LLM errors at worst fall back to code-block literals — can't inject malicious code. See AINP §17.

Q: Should I use ChatGPT / Claude / Gemini?

All 3 major LLMs support function calling / JSON mode. Accuracy (5-shot examples + function calling) roughly:

  • Claude Sonnet 4.6+: ~99%
  • GPT-4 Turbo / GPT-4o: ~98%
  • Gemini 2.5 Pro: ~97%

The differences are smaller than the impact of engineering practices (prompt design + few-shot selection + validation pipeline). Choose whichever you have API integration with.

Q: I want to fine-tune my own model — is there training data?

No ready-made fine-tuning dataset (community contributions welcome). Suggestions:

  • Collect high-quality samples from your production schema-driven generation
  • At least 500-1000 examples to start fine-tuning
  • All major LLM platforms support (OpenAI / Anthropic / Together)

3. Integrators

Q: I have my own web project and want to add Rho rendering — how do I start?

→ Developer Reference Quick start — 30-line unified pipeline integration → Install — npm install + dependencies → API reference — full API surface

Q: How do I integrate with React / Vue / Svelte?

→ Quick start framework section — per-framework patterns

Q: I'm writing a plugin for Obsidian / VS Code / Bear — how do I integrate?

Two choices:

  1. Use @rho/md library: JS projects npm install directly
  2. Implement from spec: avoid PolyForm Noncommercial license restrictions → see Implementing your own reader

Q: How do I do SSR + hydration?

→ Hydration utilities §SSR

Q: Can I do server-side rendering (not in browser)?

✅ Build-time pipeline can fully run in Node (generates HTML). ⚠️ Hydration must run in browser. SSR + client hydration pattern.

Q: How do I customize CSS?

→ CSS theming — CSS variables + class hooks + Tailwind / Emotion integration

Q: Nested DSL doesn't work — what do I do?

99% likely you forgot registerAllInnerPlugins(). See Inner processor §debugging.

Q: Is the package name stl-md or @rho/md?

The library is not published to npm yet — neither stl-md nor @rho/md resolves on the registry today. All import path docs use @rho/md as forward-looking reference.


Q: My company wants to integrate Rho into our product — do we need to pay?

See Free use FAQ decision matrix.

Simplified:

  • For-profit company using @rho/md library in production → ✅ needs commercial license
  • Self-implementing reader (not using @rho/md) → ❌ fully free (spec is CC BY 4.0)

Q: How do I get a quote?

📧 [email protected] — reply within 24h. See Commercial licensing.

Q: What's the license scope?

  • Spec (AINP): CC BY 4.0 — anyone, any purpose, free
  • @rho/md library: PolyForm Noncommercial 1.0 — commercial use requires payment
  • Rho desktop app: free for personal / commercial requires license
  • cloud.rho.md: SaaS subscription (free + pro / team / enterprise)

See License & Commercial overview.

Q: We have compliance requirements (GDPR / SOC 2 / ISO 27001) — supported?

→ DPA (Data Processing Agreement) signable → SOC 2 / ISO 27001 planned for 2026-Q4 → HIPAA / PCI not currently supported

See Commercial licensing §Legal/Compliance.

Q: Can a startup afford it?

✅ Startup-friendly pricing (< 10 people + pre-Series-B → 30-50% off). See Commercial licensing §Pricing/Discounts.


5. Protocol implementers

Q: I want to build a new reader implementation — where do I start?

→ AINP v1 protocol document — authoritative protocol definition → Implementing your own reader — implementation walkthrough

Q: What's the protocol license?

AINP is CC BY 4.0 — you can implement freely; your implementation's license is yours to choose (MIT / Apache / commercial / closed). Just preserve attribution in your docs.

Q: How much work is implementation roughly?

About 13 person-days for Full conformance. See Implementing your own reader §Recommended steps.

Q: How do I declare my reader "implements AINP"?

Per AINP §19, declare which level: Full / Static-only / Reader (no animation) / Partial.

Q: Trademark / naming rules?

✅ "MyReader (implements AINP v1)" ✅ "FooMD with AINP support" ❌ "AINP Reader" (protocol name ≠ implementation name) ❌ "Rho Clone" (Rho is the reference impl name)

See Implementing your own reader §Naming.

Q: Can I add capabilities outside the protocol?

✅ Yes ("non-standard extension"), but MUST:

  • Use distinct lang label / directive name (avoid conflicts)
  • Clearly mark in docs as "non-standard"
  • Remain readable in fallback readers

See AINP protocol.


6. Academic

Q: How do I cite AINP in a paper?

@misc{ainp2026,
  title     = {AI-native Interaction Protocol (AINP) v1},
  author    = {scos-lab},
  year      = {2026},
  publisher = {scos-lab},
  url       = {https://rho.md/protocol/RENDER_IR_v1.md},
  note      = {Working Draft. CC BY 4.0.}
}

Q: Is the Rho library free for academic research?

✅ Academic research / teaching is free (PolyForm NC explicitly includes). See Free use FAQ §Academic.

Q: Does the protocol have a reference implementation test suite?

⏸️ Conformance test pack planned for v1.0 stable release (not v1 working draft). For now, Examples gallery serves as informal reference.

Q: Are there academic papers comparing with markdown / MDX / MyST?

⏸️ No official paper yet. Community papers / blog posts welcome — can use Comparison as a starting point. If published, PRs to References section welcome.


General questions

Q: What's the relationship between AINP and Rho?

  • AINP = open protocol (CC BY 4.0)
  • Rho = scos-lab's reference implementation (npm: @rho/md, desktop app, cloud SaaS)
  • Anyone can build their own AINP-conformant reader (need not be called Rho)

See Glossary §AINP + Glossary §Rho.

Q: Where can I find an index of all docs?

Q: My question isn't here — where do I ask?

By question type:

See Community.

Q: Can I contribute docs / code?

✅ Welcomed. See Community §Contributing.


See also